Two weeks ago we updated our VRay from 2.4 to 3.0 and did some rendertests.
We currently have a projekt with mostly reflective and refractive
materials and hoped to get a speedup as we read about the raytrace speed
improvements.
But the rendertimes were twice as much as it was before on VRay2.4
There were even a difference in brightnes on two of the materials.
Our systems are
- Windows 7 64Bit, 16GB Ram
- 3ds Max 2014 (maxversion #(16000, 42, 0)
- Bitmap IN and Output Gamma is set to 1.0
- System Units are metric, Centimeters
Rendersettings and testscene are attached.
VRay 2.4 rendered in 1:25 sec and VRay 3.0 took 2:28 sec.
We already opened a support ticket[#814791] , but the developers are overloaded around SIGGRAPH.
So, maybe someone in the forum can test this on different machines and can give us a hint ?
Reset everything, vray 3.0 works differently to 2.4 and they’ve changed some of the code, especially regarding reflections. The new 2.5 build is out for those cases where you need an exact match with 2.4 but I’d suggest that you ignore your old settings. They’ve changed the sampling a little bit so when you use settings that work for vray 3, you’ll get the benefits.
It does take some getting used to, but I have to agree. Once you get the hang of it, it’s pretty great. I’ve been able to get HD frames with motion blur at 1/100 in under 15 minutes. Took a few months to really get my head around it.
The manual is always a good place to start
Forum search is another.
Try the general section (not the problems one) for the fun of it, there is a wealth of information there.
There aren’t any really specific articles talking about it, I just noticed when comparing the maya 2.4 versus 3 beta that reflections were being handled a bit differently - the chaos support staff told me in one discussion that the new vray reflections were more accurate. Likewise vlado’s changed the raytracing code a little bit so while it won’t give you the exact same results as 2.4 (it can’t since it’s different code) it’ll give you better results quicker.
for all BRDF models, John?
Aside from GGX being a special case, and the general sampler behaviour relying on the roussian roulette technique a bit more heavily than in the past, i do not remember one single beta post that talked about reflections being handled differently.
Could you elaborate?
Was just phong in a complicated blend - I’ve found the odd parameter not loading as it was in 2.4 also, things like fast sss2 would change from prepass mode in a 2.4 scene into raytraced when loaded in 3. I’ll try and dig up the reflection bits!
